Improving water extent monitoring of Swedish wetlands with open-source satellite data and Google Earth Engine
Abstract
Wetlands are strategic ecosystems that provide a variety of services to humans and the environment. As climatic and anthropogenic drivers have affected wetlands in recent years, there is a need to monitor this valuable ecosystem better. In Sweden, as a country whose fifth of its land area is covered by wetlands, water delineation is vital to understand wetlands water availability changes and strengthen their resilience to the effects of human activities and climate change. Recent advances in remote sensing have improved the monitoring of wetlands; however, mapping their water extent is still challenging. Water below vegetation is difficult to recognize, and its changes remain hidden. Here, we used different polarizations of SAR and optical data of 10 m ESA Sentinel-1&2 satellites. We processed this data in the cloud computing platform of Google Earth Engine (GEE) and took advantage of machine learning clustering methods to detect the increased backscatter due to the double-bounce of the radar signal on flooded vegetation. We also analyzed the interferometric coherence increase as a sign for inundated vegetation. Furthermore, we used the thresholding method in the SAR images to extract open water extent. Finally, we compared the results to high-resolution optical images and field data. Our workflow found areas with water below vegetation not recognized by optical methods, improving the water extent delineation in Swedish wetlands. Hence, we recommend combining polarimetric features of Radar data, optical data, and interferometry to fully account for wetland and water extent, improving the quantification of the world's surface waters.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2021
- Bibcode:
- 2021AGUFM.G51A..09M